I'm doing an RnD experiment and trying to set up dagger behind mitmproxy. I exported HTTP_PROXY and HTTPS_PROXY, added the CA cert to /etc/ssl/certs, now apt, curl and other utilities seem to be working. However, if I try to fire up dagger-anything, e.g. dagger run uptime, I'm getting this weird error message:
/ # dagger run uptime
✘ connect 0.0s
! start engine: no fallback container found
✘ starting engine 0.0s
! no fallback container found
14:44:50 WRN failed to resolve image; falling back to leftover engine error="Get \"https://registry.dagger.io/v2/\": malformed HTTP status code \"protocol\""
14:44:50 WRN failed to list containers error="exec: \"docker\": executable file not found in $PATH"
Error: start engine: no fallback container found
I am pretty confident it's got nothing to do with docker, because I can replicate it in a bare alpine:latest OCI container (docker run --rm -it --network host alpine:latest sh) by
- installing dagger (the install.sh script)
- exporting http_proxy and other similar envs, pointing to my mitmproxy instance on host (note the
--network host) and installing the mitmproxy CA cert (IDK if necessary) - running
dagger run uptime
If I unset those *_proxy envs, I can get past that error to the point where the lack of docker becomes the problem.
/ # all_proxy= http_proxy= https_proxy= ALL_PROXY= HTTP_PROXY= HTTPS_PROXY= dagger run uptime
✘ connect 1.0s
! start engine: failed to pull image: failed to run command: exec: "docker": executable file not found in $PATH
<...>
14:49:43 WRN failed to list containers error="exec: \"docker\": executable file not found in $PATH"
Error: start engine: failed to pull image: failed to run command: exec: "docker": executable file not found in $PATH
/ #
Why is this error happening? How do I get around it?